[Grok-dev] Re: grokcore.component into plone

[Philipp von Weitershausen]

Lennart Regebro wrote:
> On Fri, May 16, 2008 at 12:30 PM, Godefroid Chapelle
> <gotcha at bubblenet.be> wrote:
>> grokcore.security
> 
> I still don't think that will be reusable for Zope2.

It depends on what grokcore.security is supposed to do. For now I think 
it would only contain the grok.Permission and grok.Role base classes as 
well as the grok.require() directive/decorator. These are not tied to 
Zope 3. In fact, these are just *used* by grokkers which are indeed 
Zope3- or Zope2-specific.

> Five's security implementation is very different from Zope3s. I think
> it is best to start implementing this, and see what, if any, gets
> reusable between versions.

Exactly. I understand that the driving factor would moving things out of 
Grok is the reuse in Zope 2. However, I wouldn't want to create those 
packages of outsourced materials just so that they fit Zope 2's purpose. 
I think they should be packaged so that it makes sense for Grok and the 
rest of Zope 3.

>> grokcore.view

If this will just contain browser components (e.g. grok.View, 
grok.ViewletManager, grok.Viewlet, etc.), I'd prefer calling this 
grokcore.browser (which is more inline with Zope 3 naming conventions).

> That could work, if we make the baseclases pluggable, as Five uses
> different base classes from Zope3.

As I've said, I don't want to create reusable bits and pieces tailored 
to Zope 2. If they're reusable in Zope 3, then that's good enough for 
me. Then Five can worry about how to make it work in Zope 2. That's how 
it's always been and it has worked well so far :).

Note that the baseclass issue is completely gone on the Zope 2 trunk 
(yet another good reason for not butchering up Grok to make it fit Zope 
2). The sooner we get to release that, the sooner this problem goes away 
altogether. The only significant difference would be view security, but 
we could probably create some hooks for that.